Market-Direction-Predictor 0.1.0__py3-none-any.whl → 0.2.0__py3-none-any.whl

Market Direction Predictor/mdp.py

@@ -0,0 +1,149 @@
+from sklearn.neural_network import MLPClassifier
+from sklearn.metrics import accuracy_score,confusion_matrix
+import numpy as np
+
+def percent_change(data):return [100.0*(data[i]-data[i-1])/(data[i-1]) for i in range(1,len(data))]
+
+class MDPC:
+    """use source closing price to predict next closing price UP/DOWN"""
+    def __init__(self,clf:MLPClassifier=MLPClassifier(),wl=7):
+        self.clf=clf
+        self.wl=wl
+    def train_predict(self,closing:list[float]):
+        """trains the model based on %change of closing. return tuple of accuracy,confusion matrix,next prediction probability"""
+        data=percent_change(closing)
+        assert len(data)==len(closing)-1
+        xdata=[]
+        ydata=[]
+        T=len(data)
+        for i in range(T-self.wl):
+            current_window=data[i:i+self.wl]
+            next_value=data[i+self.wl]
+            xdata.append(current_window)
+            ydata.append(int(next_value>0.0))
+        
+        self.clf.fit(xdata,ydata)
+        ypred=self.clf.predict(xdata)
+        accuracy=accuracy_score(ydata,ypred)
+        cm=confusion_matrix(ydata,ypred)
+
+        lastwindow=data[T-self.wl:T]
+        predprob=self.clf.predict_proba([lastwindow])
+        return accuracy,cm,predprob[0]
+
+class MDPHLC:
+    """use source HLC to predict next closing price UP/DOWN"""
+    def __init__(self,clf:MLPClassifier=MLPClassifier(),wl=7):
+        self.clf=clf
+        self.wl=wl
+    def train_predict(self,high:list[float],low:list[float],closing:list[float]):
+        """trains the model based on %change of HLC data. return tuple of accuracy,confusion matrix,next prediction probability"""
+        assert len(high) == len(low) == len(closing), "All input lists must have the same length."
+        data=np.array([
+            percent_change(high),
+            percent_change(low),
+            percent_change(closing),
+        ])
+        assert data.shape==(3,len(closing)-1)
+        xdata=[]
+        ydata=[]
+        T=data.shape[1]
+        for i in range(T-self.wl):
+            current_window=data[:,i:i+self.wl].flatten(order="F")
+            next_close=data[2,i+self.wl]
+            xdata.append(current_window)
+            ydata.append(int(next_close>0.0))
+        
+        self.clf.fit(xdata,ydata)
+        ypred=self.clf.predict(xdata)
+        accuracy=accuracy_score(ydata,ypred)
+        cm=confusion_matrix(ydata,ypred)
+
+        lastwindow=data[:,T-self.wl:T].flatten(order="F")
+        predprob=self.clf.predict_proba([lastwindow])
+        return accuracy,cm,predprob[0]
+
+class MDPMORE:
+    """use multiple source/indicators (maybe volume) to predict next closing price UP/DOWN"""
+    def __init__(self,clf:MLPClassifier=MLPClassifier(),wl=7):
+        self.clf=clf
+        self.wl=wl
+    def train_predict(self,sources:list):
+        """trains the model based on %change of sources. return tuple of accuracy,confusion matrix,next prediction probability. note that sources is a list of list[float]. the last list in sources is the source that needs to be predicted(closing price)"""
+        closing_length = len(sources[-1])
+        for src in sources:
+            assert len(src) == closing_length, "All sources must match the length of the closing prices."
+        data=np.array([
+            percent_change(src) for src in sources
+        ])
+        assert data.shape==(len(sources),len(sources[-1])-1)
+        xdata=[]
+        ydata=[]
+        T=data.shape[1]
+        for i in range(T-self.wl):
+            current_window=data[:,i:i+self.wl].flatten(order="F")
+            next_close=data[-1,i+self.wl]
+            xdata.append(current_window)
+            ydata.append(int(next_close>0.0))
+        
+        self.clf.fit(xdata,ydata)
+        ypred=self.clf.predict(xdata)
+        accuracy=accuracy_score(ydata,ypred)
+        cm=confusion_matrix(ydata,ypred)
+
+        lastwindow=data[:,T-self.wl:T].flatten(order="F")
+        predprob=self.clf.predict_proba([lastwindow])
+        return accuracy,cm,predprob[0]
+
+if __name__=="__main__": # simple tests
+    d=[1,2,3 ,2,3,4 ,3,4,5 ,4,5,6]
+    mdpc=MDPC(wl=2,clf=MLPClassifier())
+    ac,cm,npp=mdpc.train_predict(d)
+    print(ac)
+    print(cm)
+    print(npp)
+    assert npp[0]>npp[1]
+    d=[1,2,3 ,2,3,4 ,3,4,5 ,4,5]
+    mdpc=MDPC(wl=2,clf=MLPClassifier())
+    ac,cm,npp=mdpc.train_predict(d)
+    print(ac)
+    print(cm)
+    print(npp)
+    assert npp[0]<npp[1]
+    print("----------------------------------")
+    h=[1,2, 1,2, 1,2]
+    l=[2,1, 2,1, 2,1]
+    c=[3,2, 3,2, 3,2] # next is 3
+    mdphlc=MDPHLC(wl=2)
+    ac,cm,npp=mdphlc.train_predict(h,l,c)
+    print(ac)
+    print(cm)
+    print(npp)
+    assert npp[0]<npp[1]
+    h=[1,2, 1,2, 1,2, 1]
+    l=[2,1, 2,1, 2,1, 2]
+    c=[3,2, 3,2, 3,2, 3]
+    mdphlc=MDPHLC(wl=2)
+    ac,cm,npp=mdphlc.train_predict(h,l,c)
+    print(ac)
+    print(cm)
+    print(npp)
+    assert npp[0]>npp[1]
+    print("----------------------------------")
+    v=[1,2,3,4, 3,4,5,6,   5,6,7,8, ]
+    c=[4,3,2,1, 3,2,1,.5, 2,1,.5,.25, ] # pred should give 1
+    mdpmore=MDPMORE(wl=3)
+    ac,cm,npp=mdpmore.train_predict([v,c])
+    print(ac)
+    print(cm)
+    print(npp)
+    assert npp[0]<npp[1]
+    h=[1,2, 1,2, 1,2]
+    l=[2,1, 2,1, 2,1]
+    c=[3,2, 3,2, 3,2] # next is 3
+    mdphlc=MDPMORE(wl=2)
+    ac,cm,npp=mdphlc.train_predict([h,l,c])
+    print(ac)
+    print(cm)
+    print(npp)
+    assert npp[0]<npp[1]

{market_direction_predictor-0.1.0.dist-info → market_direction_predictor-0.2.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: Market-Direction-Predictor
-Version: 0.1.0
+Version: 0.2.0
 Summary: simple python code for predicting markets
 Author: navidpgg
 Description-Content-Type: text/markdown

market_direction_predictor-0.2.0.dist-info/RECORD

@@ -0,0 +1,6 @@
+Market Direction Predictor/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+Market Direction Predictor/mdp.py,sha256=7mlcAoAst5ZVsYmsTfOT52GYICluJymEgHf2HSvySP8,5470
+market_direction_predictor-0.2.0.dist-info/METADATA,sha256=s2ZHxPJMMMZIgs_YJaDO_B8hfGIe_mx_xlNShkaJMPI,2507
+market_direction_predictor-0.2.0.dist-info/WHEEL,sha256=lTU6B6eIfYoiQJTZNc-fyaR6BpL6ehTzU3xGYxn2n8k,91
+market_direction_predictor-0.2.0.dist-info/top_level.txt,sha256=Pehsxxvg6ASwGJufhAl3e1ZWJqlYeU9xEv6KMVfO68g,27
+market_direction_predictor-0.2.0.dist-info/RECORD,,

market_direction_predictor-0.2.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+Market Direction Predictor

market_direction_predictor-0.1.0.dist-info/RECORD

@@ -1,4 +0,0 @@
-market_direction_predictor-0.1.0.dist-info/METADATA,sha256=mwSnOx_cNlESSkHoA_61-zLIA9mBJKX_he3nI7d9idw,2507
-market_direction_predictor-0.1.0.dist-info/WHEEL,sha256=lTU6B6eIfYoiQJTZNc-fyaR6BpL6ehTzU3xGYxn2n8k,91
-market_direction_predictor-0.1.0.dist-info/top_level.txt,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-market_direction_predictor-0.1.0.dist-info/RECORD,,

market_direction_predictor-0.1.0.dist-info/top_level.txt

@@ -1 +0,0 @@
-